Finding. A gut bacterium can steer a worm's walk by putting a siderophore on mitochondrial enzymes. Wu, Brown, Cabreiro and colleagues screened 3,985 non-essential Escherichia coli gene deletions against Caenorhabditis elegans, scored 768 behavioral features per clone, and watched an iron-homeostasis module fall out. Those mutants overproduce enterobactin in a tricarboxylic acid cycle-dependent way. Delete enterobactin biosynthesis and the movement phenotype dies. The host side is mitochondrial: the mitochondrial unfolded protein response is required for the neuromuscular, lifespan, and healthspan effects. Thermal proteome profiling puts the small molecule on mitochondrial glutaryl-CoA dehydrogenase and on electron-transfer flavoproteins A and B. Neuronal mitochondrial and cytosolic reactive oxygen species, tuned with superoxide dismutase and catalase, then gate the whole program through the secretin-family neuropeptide receptor PDFR-1.
Why this paper matters
Microbiome-brain papers often stop at "metabolite changes behavior." This one names the bacterial gene module, the metabolite, the mitochondrial proteins, the organelle stress pathway, the redox gate, and the G-protein-coupled receptor. That is a full circuit, even if it is a worm circuit. Enterobactin is not being sold as nutrition. It is being sold as a host-active iron chelator that has no business looking like a neurotransmitter and yet behaves like one at mitochondria.
GCDH and the ETF proteins are not random sticky proteins. Glutaryl-CoA dehydrogenase oxidizes a lysine and tryptophan catabolite; ETF shuttles electrons from several flavoprotein dehydrogenases into the respiratory chain. Occupying those handles is a direct way for a microbial metabolite to change mitochondrial function in neurons.
What they actually measured
The screen is large enough to be believed: Keio library coverage of non-essential E. coli genes, quantitative behavior, a coherent iron module rather than a single pretty mutant. Bacterial omics tie iron stress to siderophore output and to the TCA cycle. Loss-of-biosynthesis genetics are the necessity test. Host work combines transcription, functional mitochondrial unfolded protein response genetics, thermal proteome profiling, and redox genetics that terminate at PDFR-1. The authors' last sentence, that enterobactin is a non-canonical neurotransmitter, is a metaphor with a binding map underneath it, not a slice showing vesicular release.
How to read the score
Low 90s. Mitochondrial targets plus UPRmt plus behavior plus aging, with a screen, not a single anecdote. Confidence is high for the worm-bacterium circuit. It is low for a human gut-brain therapeutic.
Caveats
Worms. Binding is not a finished enzyme mechanism. Healthspan in C. elegans is not cognition in a person. Iron chelation can also starve the host; the paper's claim is more specific than "iron is bad," and you should keep it that way.
What to do with it
If you track microbial metabolites that touch mitochondria, put enterobactin on the same shelf as other gut-derived electron-transfer interferents. Pull the GCDH/ETFA/ETFB thermal-proteome hits and the PDFR-1 redox gate. If you work on glutaric aciduria or ETF deficiency, this is a curious environmental ligand, not a treatment hint. Do not recommend siderophore probiotics from this brief.